Method for tunable inter domain egress selection

Abstract

A flexible mechanism and method for routers to select the egress point for each destination comprises identifying a plurality of points of egress from an autonomous system, ranking the plurality of points of egress according to a metric having variable and fixed terms, selecting a point of egress having the smallest rank, and transmitting packets from a point of ingress via a path to the selected point of egress. The metric is across a plurality of destinations and respective possible points of egress from the autonomous system and the metric is m(i, p, e) equaling α(i, p, e)·d(G,i,e)+β(i, p, e) where a and β are configurable values, i is the identity of the router, p is the destination, G is an undirected weighted graph, the d function is the interior gateway protocol distance and e is a point of egress.

Description

BACKGROUND OF THE INVENTION[0001]The invention relates to the field of data transmission protocols and, more particularly, to the field of selecting an efficient point of egress from a data network by ranking possible points of egress according to principles of tunable inter-domain egress (TIE) as further explained herein.[0002]The Internet's two-tiered routing architecture was designed to have a clean separation between intra-domain and inter-domain routing protocols. For example, an inter-domain protocol allows the border routers to learn how to reach external destinations, whereas the intra-domain protocol determines how to direct traffic from one router in an autonomous system (AS) to another router. However, the appropriate roles of the two protocols becomes unclear when the autonomous system learns routes to a destination at multiple border routers—a situation that arises quite often today. An autonomous system as defined by Newton's Telecom Dictionary is a collection of route...

AI Technical Summary

Benefits of technology

[0021]In a preferred embodiment of the present invention, candidate egress points are compared based on a weighted sum of the IGP distance and a constant term. The configurable weights provide flexibility in deciding whether (and how much) to base BGP decisions on the IGP metrics. Network management systems may apply optimization techniques to automatically set these weights to satisfy network-level objectives, such as load balancing objectives and the minimization of propagation delays. To ensure consistent forwarding through the network, lightweight tunnels are used to direct traffic from the ingress router to the chosen egress point. The new preferred method, according to the present invention, will be referred to herein as TIE (Tunable Inter-domain Egress) because it controls how routers break ties between multiple equally-good BGP routes. TIE is both simple (for the routers) and expressive (for network administrators). No new protocols or any changes to today's routing protocols are introduced with TIE, making it possible to deploy at one autonomous system at a time and with only minimal changes to the BGP decision logic on IP routers. Thus, one aspect of the present invention is to provide a mechanism for egress-point selection that is flexible enough to control the flow of traffic in steady state, while responding automatically to network events that would degrade performance.

[0022]In accordance with one aspect of the present invention, TIE provides a flexible mechanism for egress-point selection. In particular, the TIE mechanism is: (i) flexible in balancing the trade-off between sensitivity to IGP changes and adaptability to network events, (ii) computationally easy for the routers to execute in real time, and (iii) easy for a higher-level management system to optimize based on diverse network objectives.

[0023]In accordance with another aspect of TIE, network-wide objectives may be met. Exemplary problems are suggested and are solved easily using TIE. First, sensitivity to internal topology changes is minimized, subject to a bound on propagation delay, using integer programming to tune the weights in the TIE mechanism. Second, load is balanced in the network without changing the IGP metrics or BGP policies, by using multi-commodity-flow techniques to move some traffic to different egress points than would be selected using prior art techniques.

[0024]In particular, TIE has been evaluated on two backbone networks, Abilene and a large tier-1 ISP, using traffic, topology and routing data from these two backbone networks. Results indicate that TIE reduces sensitivity to internal topology changes while satisfying network-wide objectives for load and delay.

Problems solved by technology

However, the appropriate roles of the two protocols becomes unclear when the autonomous system learns routes to a destination at multiple border routers—a situation that arises quite often today.

In addition, hot-potato routing tends to limit the consumption of bandwidth resources in the network by shuttling traffic to the next autonomous system as early as possible.

The decision to select egress points based on IGP distances may be inappropriate in light of the growing pressure to provide good, predictable communication performance for applications such as voice-over-IP, on-line gaming, and business transactions.

Hot-potato routing may be unnecessarily restrictive.

Moreover, “hot potato” routing tends to be disruptive.

Tying egress selection to IGP distances may lead to harmful disruptions and over-constrained traffic-engineering problems.

Also, allowing each ingress router to have a fixed ranking of egress points may not be flexible enough (for traffic engineering) or adaptive enough (to large changes in the network topology).

Under hot-potato routing, point of ingress router C into AS 1 chooses the BGP route learned from A because the IGP distance to A is 1+1 or 2, which is smaller than the distance of 9 to B. However, if the C-D link fails (indicated by the X break), all traffic from ingress C to destination p would shift to egress router B, with an IGP distance of 9 that is smaller than the IGP distance of 10 to alternative egress router A. These kinds of routing changes are disruptive.

Yet, continuing to use egress-point A might not be the right thing to do, either, depending on the propagation delay, traffic demands, and link capacities.

Although hot-potato routing is a reasonable way to minimize resource consumption, IGP link weights do not express resource usage directly.

The IGP distances do not necessarily have any relationship to hop count, propagation delay, or link capacity, and selecting the closer egress point does not necessarily improve network performance.

In addition, small topology changes can lead to performance disruptions, for example, large shifts in traffic within and between autonomous systems.

A single link failure can potentially impact the egress-point selection for tens of thousands of destinations at the same time, leading to large shifts in traffic.

Another type of performance disruption is changes in the downstream path.

When the egress point changes, the traffic moves to a different downstream forwarding path that may have a different round-trip time or available bandwidth, which may disrupt the communicating applications.

In addition, the abrupt increase in traffic entering the neighboring AS may cause congestion.

Yet another performance disruption is the need for BGP update messages for neighboring domains.

Even if the hot-potato routing change does not lead to new BGP update messages, long convergence delays can occur inside the autonomous system depending on how the router implements the BGP decision process.

Long convergence delays may occur because the underlying routers in the network only revisited the influence of IGP distances on BGP decisions once per minute; during the convergence period, data packets may be lost, delayed, or delivered out of order.

Not all of these events are caused by unexpected equipment failures—a large fraction of them are caused by planned events, such as routine maintenance performed by service personnel.

Besides being disruptive, the tight coupling between egress selection and IGP metrics makes traffic engineering and maintenance planning extremely difficult.

However, finding good settings that result in the desired behavior is computationally challenging, due to the large search space and the need to model the effects on egress-point selection.

Finding settings that are robust to a range of possible equipment failures is even more difficult, imposing even more constraints, such as minimizing hot-potato disruptions across all routers and destination prefixes and making the problem increasingly untenable.

In addition, once local-search techniques identify a better setting of the IGP metrics or BGP policies, changing these parameters in the routers requires the network to go through routing-protocol convergence, leading to transient performance disruptions.

Hot-potato routing adapts immediately to internal routing changes (however small), leading to frequent disruptions.

Imposing a fixed ranking of egress points, while robust to topology changes, cannot adapt in real time to critical events.

Neither mechanism offers sufficient control for network administrators trying to engineer the flow of traffic and plan for maintenance.

Images